Speed handicapping - a mystery

[cosmicway]

How do we compute the win probabilities of the horses in a given card in a speed model ?
The common approach is this:

- compute the expected speed for each horse using histogram analysis
- compute a function of the form p ~ exp(-(t - t0)/τ) and normalise to 1.

In this formula t is the horse's expected speed (time from wire to wire), t0 is the best expected speed in the group and τ is a time constant characteristic of the distance.

There are more elaborate ways, including the use of a multiple integral to compute the p values, but to a first approximation this is it.

Now suppose we want the probabilities for seconds and thirds, so we can go for the forecasts and tricasts.
How to do this ?

We use Harville's law.
This in bare form is grossly inaccurate but we can make corrections.
First the histogram analysis has to be modified, second the time constant is not the same.

After all this work suppose:

We find a mean probability Q1 for the winners.
We find for the second finishers a mean probability Q2, this being the mean probability with the winners supposedly known (before we add the lost using Harville to work out our p's).
We find for the third finishers a mean probability Q3, this being with the winners and the second finishers supposedly known.

We expect that Q3 > Q2 > Q1.
Because for the second finishers we have to deal with N-1 horses and for the third finishers with N-2, fewer that is.

But in all the databases I have analyzed it turns out that it is not so.
Q2 is less than Q1 !
Q3 is better than Q1 and Q2 by a margin (this being as expected).

Because the same thing was observed everytime I did this kind of analysis, it cannot be chance. It's a property of nature (horse racing nature).
Why you think ?

[sjk]

I don't know Harvilles Law but I make probability curves based on projected speed figures and then do a calculation of the likelihood of each horse winning based on the naive assumption that the curves are independent. The result is useful but needs a lot of adjustment as you point out.

Speed horses will outperform and closers will underperform for the win spot. This is less true for the lower placings.

[cosmicway]

Originally Posted by sjk

I don't know Harvilles Law but I make probability curves based on projected speed figures and then do a calculation of the likelihood of each horse winning based on the naive assumption that the curves are independent. The result is useful but needs a lot of adjustment as you point out.

Speed horses will outperform and closers will underperform for the win spot. This is less true for the lower placings.

Yes of course.
The closers are at a handicap in sprint distances but less so for the minor placings.

But in my -partial- calculations the early speed is not factored in.
Only wire to wire.

Play this game with a friend:
He tasks you to study all the race cards of the day and then he locks you into a dark room without tv. After the races he comes in and asks you to name the winners. You register a certain score, say 25%.
This is experiment number 1 and it lasts for one thousand years.

Now play this:
Again he locks you in tha dark room but this time he announces the winners to you and you have to name the seconds.
This is experiment number 2 and it lasts for another thousand years.
But you will register a lower score, perhaps 23%.

Of course if you make random guesses then your success rate in experiment 1 will be 1/N and in experiment no 2 it will be 1/(N-1), higher.
But I mean using your tested speed model you lose in the game of seconds.

Then if you experiment more for the third placed horses given thefirst two, you will score higher. The second placed horse is the discrepancy.

I believe this happens with all speed models.

[sjk]

No need for me to play make believe games that last 1000 years. I have 30 years of history betting real money (70,000 races). I am pleased with my ability to make probabilities as I was discussing above. Unfortunately others are doing the same or better and have made the game increasingly difficult (and boring as most races are a pass).

[cosmicway]

Originally Posted by sjk

No need for me to play make believe games that last 1000 years. I have 30 years of history betting real money (70,000 races). I am pleased with my ability to make probabilities as I was discussing above. Unfortunately others are doing the same or better and have made the game increasingly difficult (and boring as most races are a pass).

The enemy is the taxman.
They said "ok, the punters are mugs - they can't do nothing" and they were right. Not only punters are mugs but many of them did n't even get to know it when the taxes went up.
But - but there is another law that says "if the money flow out of a man's pocket they 're not coming back".
So turnovers are dwindling and horse racing is marginalized as a sport and not paradoxically those who are taxing us don't see any money either.

Now as you say you have a history of 70,000 races I wonder if you have or have n't come across this thing I 'm talking about.

[sjk]

It is the computer assisted guys doing the same thing we are doing but better that are taking the money.

Of course taxes are never much fun either.

[cosmicway]

Originally Posted by sjk

It is the computer assisted guys doing the same thing we are doing but better that are taking the money.

Of course taxes are never much fun either.

I can't be certain of that, maybe. But evidently when somebody steals your weapon you no longer have the edge. Like with the A-bomb.

Now I wonder if somehow my pdf function for the places is somewhat off, but I believe there is somewhere an inner reason and its important because my type of play -and many peoples- is the quinella-forecast.

[cosmicway]

This is still a mystery.
I tweaked the pdf function - the same.
I checked with other tips providers - the same happens with their tips.

My calculations involve only wire to wire speeds, they are partial calculations in a bigger prediction model.
But issues relating to early positions should not enter that being so.

Nevertheless I propose a certain explanation which may or may not be true:
In the majority of the races the top favourites are among the front runners.
One of them often wins but because of the speed duels the others perish.
Therefore the second finisher tends to be from all over the place, thus making the pdf for the seconds score lower.

[46zilzal]

Mystery? You must be kidding

[cosmicway]

Originally Posted by 46zilzal

Mystery? You must be kidding

Why am I kidding ?
I don't know about you but my play is forecast-quinella-trio, win less often.
That's why I 'm into the details.
You have some other explanation ?